[ExI] Why Cities Keep Growing, Corporations and People Always Die, and Life Gets Faster

[Anders Sandberg]

I recommend his paper "Growth, innovation, scaling, and the pace of life 
in cities" at
http://www.pnas.org/content/104/17/7301.short
(free access) It gives some of the data and modelling, as well as the 
singularity model he gets.

Overall, it seems to be a "law" that if anything has a superlinear 
factor in its growth, it tends to have a finite time singularity.

Kelly Anderson wrote:
> The discussion of the Singularity is that it can be avoided for a
> while by using technology to push it off. I think this is a different
> sort of singularity than what we are used to talking about here. This
> is a singularity in the growth of a particular city or company, not of
> mankind in general, although that could be interpolated from what he's
> saying by treating the entire planet as a biological system of the
> type he's discussing.
>   

Yup. His model is essentially arguing that we will get a finite time 
singularity that runs out of resources. A correlate is that if we want 
to have a "sustainable" singularity we better have our growth in domains 
that are more and more resource efficient.

Note that the other kinds of singularities we often talk about 
(superintelligence, prediction horizons, intelligence explosions, phase 
transitions, ...) may or may not have resource limits.

>> You need major innovation on smaller time
>> scales to continue growing the way we do.
>>     
>
> Yes, this is where he falls off the rails. He dismisses it as
> unrealistic that human beings can incorporate major new technologies
> on the time scale of 6 months. Well, perhaps human beings can't, but
> humanity+ can, and probably will. AGI can and probably will. I don't
> think he is aware of the Singularity that we are all familiar with, or
> at least he doesn't address that at all in this discussion. It would
> be fascinating to discuss it with him.
>   

He is kind of aware. I talked with him about it, and he plans to meet 
with Nick next time he is over here in Oxford. It is just that he 
doesn't buy into AGI and similar things straight away, and at the very 
least not that that it would not be having the same kind of resource 
limits. Also, what happens to the AGI cluster when it needs to innovate 
faster than it can communicate the innovations to the other parts due to 
lightspeed limits?


> I wonder if he's ever applied his technique to governments? Now THAT
> would be interesting. I'd bet governments grow sublinearly at a rate
> below that of even companies, but that's just a guess. It feels like
> they get less efficient with growth, but that would contradict his
> entire thesis. This would be interesting indeed.
>   

Would bthe number of laws or government employees count as growth? In 
that case I suspect superlinear growth and the prediction of a 
bureaucratic singularity!

Quick check: http://www.data360.org/dsg.aspx?Data_Set_Group_Id=228
Ha! I was apparently wrong, the number of federal employees has 
remaining pretty constant. Local employees are increasing but perhaps 
merly linearly. I suspect the volume of laws is increasing faster, but 
you need to specify what to measure carefully.


-- 
Anders Sandberg,
Future of Humanity Institute 
James Martin 21st Century School 
Philosophy Faculty 
Oxford University